Thread: resend: Chinese sort order problem
hi all,
I am shade to disturb people here again, but to my Question , i
have not effective answer.
it has been suffering me for 3 weeks.
the problem is : How the Postgresql treet the Chinese Text Field
sort（order by）?
It seems always false sort order. but the MySQL does this very well.
in my machine "select getdatabaseencoding();" returns "EUC_CN"
and "\encoding" returns "EUC_CN" too.
I have tested for "createdb -E EUC_CN / SQL_ASII / LATIN1" ,
the option "EUC_CN" / "SQL_ASII" results same .
I have also tried recompling the source with "configure
--enable-locale --enable-multibyte=EUC_CN", it does no help.
So i want to know where is the source code of SQL "ORDER BY", if
nobody else could help me, I can only choose to
study the source and solve it by myself.
Any suggestion is kind to me.
Thanks.
Malix
Hello Malix, There are some limitations to multi-byte support. I had the same problems, see another mail I forwarded. I am new to the list but I suggest you sent this message to pgsql-hackers@postgresql.org. Regards, Jean-Michel POURE At 15:45 29/10/01 +0800, you wrote: >hi all, > I am shade to disturb people here again, but to my Question , i >have not effective answer. > it has been suffering me for 3 weeks. > > the problem is : How the Postgresql treet the Chinese Text Field >sort£¨order by£©? > It seems always false sort order. but the MySQL does this very well. > > in my machine "select getdatabaseencoding();" returns "EUC_CN" > and "\encoding" returns "EUC_CN" too. > I have tested for "createdb -E EUC_CN / SQL_ASII / LATIN1" , > the option "EUC_CN" / "SQL_ASII" results same . > > I have also tried recompling the source with "configure >--enable-locale --enable-multibyte=EUC_CN", it does no help. > > So i want to know where is the source code of SQL "ORDER BY", if >nobody else could help me, I can only choose to >study the source and solve it by myself. > > >Any suggestion is kind to me. >Thanks. >Malix > >---------------------------(end of broadcast)--------------------------- >TIP 6: Have you searched our list archives? > >http://archives.postgresql.org
> hi all, > I am shade to disturb people here again, but to my Question , i > have not effective answer. > it has been suffering me for 3 weeks. > > the problem is : How the Postgresql treet the Chinese Text Field > sort（order by）? > It seems always false sort order. but the MySQL does this very well. > > in my machine "select getdatabaseencoding();" returns "EUC_CN" > and "\encoding" returns "EUC_CN" too. > I have tested for "createdb -E EUC_CN / SQL_ASII / LATIN1" , > the option "EUC_CN" / "SQL_ASII" results same . > > I have also tried recompling the source with "configure > --enable-locale --enable-multibyte=EUC_CN", it does no help. You did not mention what is "false" and what is "correct" sort order. So I'm not sure I understand you problem, but I experience similar ones with Japanese on some Linux platforms. In my case the source of the problem was broken locale database coming with Linux. After re-building PostgreSQL WITHOUT --enable-locale, all problems were gone. Hope this helps. -- Tatsuo Ishii
hello Tatsuo,
thank you very much, your suggestion solved my problem now.
It can now "order by" with correct Chinese PinYin sort order. I am really
very happy.
Would you like to change the description in the Reference Manual for
Installation?
My Problem is a result following the Information in that Chapter, always
recompiling with "--enable-locale" for my locale support.
Maybe I have incorrect understand about the following:
--enable-locale :: for support locale language sort order
and --enable-multibyte :: for multibyte character storing ?
So as a conclusion , when people want Chinese support in Postgres, he must
recompile the source,
with --enable-multibyte=EUC_CN, but without --enable-locale.
Thanks
Malix
> hello Tatsuo, > thank you very much, your suggestion solved my problem now. > It can now "order by" with correct Chinese PinYin sort order. I am really > very happy. > Would you like to change the description in the Reference Manual for > Installation? > My Problem is a result following the Information in that Chapter, always > recompiling with "--enable-locale" for my locale support. > Maybe I have incorrect understand about the following: > --enable-locale :: for support locale language sort order > and --enable-multibyte :: for multibyte character storing ? I would say multibyte character sets sorting using locale would never work or it returns the same result without locale at best. > So as a conclusion , when people want Chinese support in Postgres, he must > recompile the source, > with --enable-multibyte=EUC_CN, but without --enable-locale. That might be a good idea. The locale support is useless anyway for multibyte character sets such as Chinese and Japanese. I'm not sure about traditional Chinese, Korean and Unicode though. Another workaround would be using "C" locale anytime. But the question is, if always using "C" locale, why you need the locale support:-) -- Tatsuo Ishii
At 12:39 30/10/01 +0800, you wrote: > --enable-locale :: for support locale language sort order > and --enable-multibyte :: for multibyte character storing ? Does --enable-multibyte set --enable-locale automatically in 7.2? Do I miss something? Regards, Jean-Michel
> At 12:39 30/10/01 +0800, you wrote: > > --enable-locale :: for support locale language sort order > > and --enable-multibyte :: for multibyte character storing ? > > Does --enable-multibyte set --enable-locale automatically in 7.2? No. --enable-multibyte just implies --enable-unicode-conversion. -- Tatsuo Ishii